What are your special interests?

My special interests are:

1. Coding - game dev with some games released on itch.

2. Math puzzles - I found a new pattern in a unsolved Math problem.

3. AI - highly obsessed - talk to AI every day learning new things!

4. Trying to learn everything! e.g i know it's impossible to learn everything ( but golly gosh I'm going to try! )

5. trying to find connections with seemly unrelated things!

I'm with you on no. 4! I've been trying to learn everything about everything since I was about 2 years old. Half a century in and I've still got a long way to go. It's what gets me out of bed every morning (unless I'm reading something so interesting in bed that I forget to get up until the afternoon). As a special interest, I think it'll see me out.

I mentioned to someone else on the forum the book Why Machines Learn by Anil Ananthaswamy. If you're interested in maths (2) and AI (3), it might float your boat. I liked it.

It's a nice feeling when people have the same special interest! - I've recently started a project programming my own AI, but I'm taking a unconventional approach to it - since as you know from reading " Why Machines Learn"
current AI uses floating-point weights -1 and 1 binary but I'm going down the root of trying to make a AI using Ternary method which uses 3 states (-1, 0 and 1), because i believe AI in it's current state can never reach AGI.

couple years ago was talking to a old friend about this, and part of me wanted to learn it from the base up and not just copy and paste a pre-build model ( but he refused to start/learn from scratch ) because my inner logic was buzzing thinking it was missing something like a color layer to it, but i was not sure till recently i came across Ternary computing from 1958! then if i get better results ( will build both versions ) then try to design a Ternary CPU, if it works out! still trying to get my head around it all!

also on the same note, today i contacted Deepmind about working with them/for them, but i do admit i lack a lot of skills necessary! but if i get this new AI built and working could always contact them again if they don't get back to me! - with more proof of my skills*

It's a nice feeling when people have the same special interest! - I've recently started a project programming my own AI, but I'm taking a unconventional approach to it - since as you know from reading " Why Machines Learn"
current AI uses floating-point weights -1 and 1 binary but I'm going down the root of trying to make a AI using Ternary method which uses 3 states (-1, 0 and 1), because i believe AI in it's current state can never reach AGI.

couple years ago was talking to a old friend about this, and part of me wanted to learn it from the base up and not just copy and paste a pre-build model ( but he refused to start/learn from scratch ) because my inner logic was buzzing thinking it was missing something like a color layer to it, but i was not sure till recently i came across Ternary computing from 1958! then if i get better results ( will build both versions ) then try to design a Ternary CPU, if it works out! still trying to get my head around it all!

also on the same note, today i contacted Deepmind about working with them/for them, but i do admit i lack a lot of skills necessary! but if i get this new AI built and working could always contact them again if they don't get back to me! - with more proof of my skills*

Zyphullen said:

3 years making games and never finished a single project or released anything on steam as i planned to, kinda why i wanted ADHD assessment

That sounds so familiar! It helps to reframe it as having had lots of fun doing it, rather than being hard on yourself for not completing it.

I agree with Trev about the WCAG. I struggled to read any text on your page. Take a look at things like minimum contrast ratios. If you google "contrast checker" you'll find lots of sites that will help you get the text contrast right for old fogies like me. If in doubt, there's nothing wrong—and a whole lot right—with black on white.

I wish you  the best of luck with your course and getting your games finished - I'm sure you will get there in the end.

Yes i have, but not when i made these games, just started the course 2 months ago, before hand was kinda just winging it, i re-invented the wheel in some of my games ( back end code wise ) making my own code versions for what is already build into Unity. 

also most games only work on desktop not mobile or tablet, and all my games are just prototypes next to no views... kinda de-motivating... but i have a issue where i have to many different ideas and struggle to stick to 1 at a time, got like 30 projects on a go and keep jumping from one to another so many unfinished projects.

3 years making games and never finished a single project or released anything on steam as i planned to, kinda why i wanted ADHD assessment / medication to keep me on track but it will never happen or it will when I'm 97 years old!

As part of your course have you covered the Web Content Accessibility Guidelines:

https://www.w3.org/TR/WCAG22/

I mention it as I struggled to read some of the text on your web page (I'm not colour blind).

I've always tried to make sure any website I created would work on any device and be readable by anyone.

same but with games, my art skills / UI skills are bad! but i love coding so same boat! maybe but i know a few autistic people who hate coding but love making things look good/art side 

https://zyphullen.itch.io/ for example 

I don't know if it is an autism thing or just a personal shortcoming, but I've always been rubbish at knowing how to make websites look nice - I could do the technical bit (what you call full stack), but my websites have always been really plain (I've never done web development professionally, but have created websites for charities I've been involved with, a couple of websites selling products and personal websites).

I'm currently doing a Full stack course with Learning People - since i have no proof of skills other then my games released on itch.io to help me through the door since I'm currently unemployed, nearly finished Front end so will be going through the back end soon.

i appreciate the advise though, but knowing me I'll stick with what i know for now! once i get a model up and running and the basics down move over to python ( but Unity's Inspector to me is like gold dust for reading data ) - old habits die hard!

You can do the CUDA stuff in Python, too. If you want a job in the AI field, putting Python on your CV should probably be high on your list of priorities.

Disclaimer: I don't code in Python, but I do work on the periphery of AI, so I know that Python is the darling of all those AI boffins.

I tried using Python while trying to solve the magic square of squares, and that inefficiency really affected it, so I ended up sticking with Unity C# due to CPU parallelization and even tried a GPU method, but the GPU method was like 1000x faster but about the same amount in difficulty ( needing to download CUDA and setup Compute shaders ) setup was a nightmare since it's not built for these types of algorithms ( crashed my GPU like 100+ times due to memory overload and leaks ) so gave up then ended up rapid prototyping so quickly it ended up being better to just stick with parallelization.

since I've used C# for 3 years it's just easier to edit/improve and maybe better since everyone uses python and if you want to innovate you need to try things in different ways, but both kinda same base language anyways... hmm maybe backtracking but either way just stick with what i know for now, can always just ask Grok to convert A script to B script if needed.

Your mention of Ternary made me think of analog computers - I had considered building one a few years ago, but had enough ongoing projects at the time so didn't.  If you are interested I found this relatively recent article (from 2019):

www.nature.com/.../s41598-019-49204-y

Zyphullen said:

current AI uses floating-point weights -1 and 1 binary

Floating-point values (including their use as weights in neural networks) represent real numbers. They are not limited to -1 and 1. FP8 and even FP4 are common in AI, as they allow more calculations to be done more quickly, while still being "good enough" to represent weights.

Any ternary CPU can be emulated with a binary CPU. Both are essentially Turing Machines, so they are, in that sense, equivalent.

If you want to have a go at building your own AI or dabbling with machine learning, I recommend coding in Python. Python has lots of nice libraries for building and configuring neural networks and such. It's the go-to language of AI these days (which is odd, given how inefficient it is).